Bromodomain and extra terminal protein (BET) inhibitors are first-in-class targeted therapies that deliver a new therapeutic opportunity by directly targeting bromodomain proteins that bind acetylated chromatin marks1,2. Early clinical trials have shown promise, especially in acute myeloid leukaemia3, and therefore the evaluation of resistance mechanisms is crucial to optimize the clinical efficacy of these drugs. Here we use primary mouse haematopoietic stem and progenitor cells immortalized with the fusion protein MLL-AF9 to generate several single-cell clones that demonstrate resistance, in vitro and in vivo, to the prototypical BET inhibitor, I-BET. Resistance to I-BET confers cross-resistance to chemically distinct BET inhibitors such as JQ1, as well as resistance to genetic knockdown of BET proteins. Resistance is not mediated through increased drug efflux or metabolism, but is shown to emerge from leukaemia stem cells both ex vivo and in vivo. Chromatin-bound BRD4 is globally reduced in resistant cells, whereas the expression of key target genes such as Myc remains unaltered, highlighting the existence of alternative mechanisms to regulate transcription. We demonstrate that resistance to BET inhibitors, in human and mouse leukaemia cells, is in part a consequence of increased Wnt/β-catenin signalling, and negative regulation of this pathway results in restoration of sensitivity to I-BET in vitro and in vivo. Together, these findings provide new insights into the biology of acute myeloid leukaemia, highlight potential therapeutic limitations of BET inhibitors, and identify strategies that may enhance the clinical utility of these unique targeted therapies.

Deep mutational scanning is a widely used method for multiplex measurement of functional consequences of protein variants. We developed a new deep mutational scanning statistical model that generates error estimates for each measurement, capturing both sampling error and consistency between replicates. We apply our model to one novel and five published datasets comprising 243,732 variants and demonstrate its superiority in removing noisy variants and conducting hypothesis testing. Simulations show our model applies to scans based on cell growth or binding and handles common experimental errors. We implemented our model in Enrich2, software that can empower researchers analyzing deep mutational scanning data.Electronic supplementary materialThe online version of this article (doi:10.1186/s13059-017-1272-5) contains supplementary material, which is available to authorized users.